BACTERIA MAY CONTRIBUTE TO DISTANT SPECIES RECOGNITION IN ANT-APHID MUTUALISTIC RELATIONSHIPS Christophe Y. Fischer1*, Claire Detrain2, Philippe Thonart3, Eric Haubruge4, Frédéric Francis4, François J. Verheggen4 and Georges C. Lognay1 1 Laboratory of Analytical Chemistry, University of Liege, Gembloux Agro-Bio Tech, Passage des Déportés 2, 5030 Gembloux, Belgium. ([email protected]; [email protected] ) 2 Unit of Social Ecology, Université Libre de Bruxelles, CP231,50 avenue F. Roosevelt, B-1050 Brussels, Belgium. ([email protected]) 3 Walloon Center of Industrial Biology, University of Liege, B40, 4000 Sart-Tilman, Belgium. ([email protected]) 4 Functional and Evolutionary Entomology, University of Liege, Gembloux Agro-Bio Tech, Passage des Déportés 2, 5030 Gembloux, Belgium. ([email protected]; [email protected]; [email protected]) *Corresponding author present address: Christophe Fischer Centre wallon de Recherches agronomiques Département Valorisation des productions Bâtiment Maurice Henseval Chaussée de Namur, 24 5030 Gembloux, Belgium [email protected] 0032 81 62 03 70 1 Abstract Mutualistic interactions between ant and aphid species have been the subject of considerable historical and contemporary investigations, the primary benefits being cleaning and protection for the aphids and carbohydrate-rich honeydew for the ants. Questions remained, however, as to the volatile semiochemical factor influencing this relationship. A recent study highlighted the role of bacterial honeydew volatile compounds in ant attraction. Here, ant’s ability to distantly discriminate two aphid species was investigated based on bacterial honeydew semiochemicals emissions using a two-way olfactometer. Both the mutualistic black bean aphid (Aphis fabae L.) and the non-myrmecophilous pea aphid (Acyrthosiphon pisum Harris) were found to be attractive for the black garden ant (Lasius niger L.). The level of attraction was similar in both assays (control versus one of the aphid species). However, when given a choice between these two aphid species, ants showed a significant preference for Aphis fabae. Honeydew volatiles, mostly from bacterial origins, are known to be a key element in ant attraction. Using the same olfactometry protocol, the relative attractiveness of volatiles emitted by honeydews collected from each aphid species and by bacteria isolated from each honeydew was in investigated. Again, ants significantly preferred volatiles released by Aphis fabae honeydew and bacteria. This information suggests that microbial honeydew volatiles enable ants to distantly discriminate aphid species. These results emphasize the importance of investigating the presence and potential effects of microbes in insect symbioses. Keywords: ant, aphid, bacteria, honeydew, mutualism, recognition, VOC 2 Introduction Ant-aphid interactions, as a major model of mutualistic relationships, have been the subject of considerable historical and contemporary investigations. Small and defenseless aphids are easy prey for numerous predators and parasitoids. However, some aphid species are frequently found in association with ants that tend and protect the aphids in exchange of honeydew, which is a reliable and abundant carbohydrate source. Other aphid species do not develop such partnerships with ants and are rather considered as preys (Dixon 1985; Bristow 1991; Stadler and Dixon 2005). In Europe, it has been estimated that myrmecophily, i.e. ants tending, is observed for two thirds of aphid species (Stadler 1997). Three ant genera appear to be primarily involved in these mutualistic interactions: Lasius, Myrmica and Formica (Stadler and Dixon 1999; Guénard et al. 2007), among which the black garden ant, Lasius niger L. (Hymenoptera, Formicidae), is well known to tend several aphid species, including the black bean aphid, Aphis fabae Scopoli (Homoptera, Aphididae) (El-Ziady and Kennedy 1956). These two species are frequently used in studies on mutualistic interactions. Besides, the pea aphid, Acyrthosiphon pisum Harris can be reared on the same host plant, Vicia faba L., as the bean aphid but is never observed in association with ants. For these reasons, these three insect species, widespread in their natural environment, were selected as model taxa for this research (Wilson 1955; Holman 2009). While aphids do not seem to actively search for ant partners, ants are known to search for their aphid partners and are able to use different volatile cues to orientate this search, namely volatile organic compounds emitted by honeydew bacteria (mVOCs) and E-β-farnesene (EBF), an aphid pheromone involved in alarm and social behaviors (Verheggen et al. 2012; Fischer et al. 2015a; Fischer et al. 2015b). This last molecule is a major component of the alarm pheromone of numerous aphid species, and is thus not likely to constitute alone a suitable species recognition cue (Francis et al. 2005). In contrast, honeydew VOCs emission profiles vary between aphid species, depending on various factors including the aphid species and its gut microflora, and could thus contribute to distant aphid discrimination. This paper describes a study designed to investigate the role of honeydew mVOCs in distant aphid discrimination by ant partners. 3 Material and methods Plants and insects In a climate-controlled room (16 h light-8h dark photoperiod; 20 ± 2°C), black bean aphids, Aphis fabae, and pea aphids, Acyrthosiphon pisum, were reared for several generations on broad beans, Vicia faba, cultivated on a 1/1 mix of perlite and vermiculite substrate. Plants used in experiments were about 15 cm high. All substrates used in behavioral assays were previously sterilized. Lasius niger colonies were collected in Gembloux (Belgium), and kept under the same climatic conditions, but in separate chambers. Nests were placed in plastic containers coated with polytetrafluoroethylene (Fluon®,Whitford, U.K.) to prevent escape. Test tubes covered with a red transparent foil were used as laboratory rearing nests; a water and aqueous brown sugar solution (342 g/L) was provided ad libitum, and dead insects (fruit flies and mealworms) were provided weekly as an additional food source. All nests used in the bioassays were comprised of a queen, brood, and a minimum of 500 foragers. Choice tests with two-way olfactometer The level of attraction on ants was assessed for different samples by using a protocol previously successfully used for similar tests with ants (Fischer et al. 2015a). The two-way olfactometer consisted of a Y-shaped glass tube (diameter: 1.5 cm, entrance length: 20 cm; length of each arm: 30 cm). Samples and controls were placed in 4 L glass jars. Filtered air was forced into the jars at 200 mL/min and delivered to the olfactometer’s branches via Teflon® tubing. Ants were starved for three days prior to an assay. The olfactometer’s entrance was placed in the ant’s rearing container, allowing the ants to enter the system. Only one ant worker at a time was allowed in the olfactometer; the entrance was closed to prevent additional ants from entering after the first ant went in. The choice, i.e. the branch selected to reach the end of the olfactometer arm, and the average linear speed while passing through the branch were recorded for each tested ant. The test ended when the ant reached a point located 25 cm from where the two arms branched (“choice point”). The linear speed was calculated by measuring the time spent to pass through an olfactometer branch, and was expressed in cm/s. The attractiveness of 4 a sample was expressed by the relative number of ants (%) choosing the particular sample side as their final choice. All assays were conducted at 20°C in a dark-walled chamber presenting no visual cues that may influence ant choices. To prevent ants from laying trails, they were never allowed to reach the actual samples. Moreover, in order to palliate any potential bias induced by the environment or by any marking of the substrate by exploring ants, sample and control sides were switched every five ants. The olfactometer was completely cleaned every 20 ants. Several sample-control couples were tested following this protocol (Table 1). For each modality, samples and controls presented to ants were used to test 20 ants and then renewed.. The first samples were aphid-infested plants (a pot holding nine V. faba infested for three days either by 50 myrmecophilous Aphis fabae or by 50 non-myrmecophilous Acyrthosiphon pisum respectively); their attractiveness were compared to that of a healthy plant without aphids. The following test assessed the relative attractiveness of the two first samples, plants infested by one of these two species being presented at each side of the olfactometer. The global attractiveness of an aphid-infested plant relies mostly on volatile organic compounds (VOCs) released by the honeydew accumulating around aphid colonies, and more specifically by aphid-associated bacteria present in honeydew (Fischer et al. 2015a). The relative attractiveness of honeydews was thus assessed for the two tested aphid species. Sample consisted of Aphis fabae honeydew that was collected for three days from a heavily infested plant onto wet substrate (perlite/vermiculite 1:1) to avoid desiccation, while control was Acyrthosiphon pisum honeydew (collected the same way). The relative attractiveness of two taxonomically close honeydew bacteria known to be involved in aphid interactions with other insect species was also tested. The first one, Staphylococcus xylosus, is found in Aphis fabae gut and honeydew and is known to produce mVOCs attractive for L. niger (Fischer et al. 2015a); the second one, Staphylococcus sciuri, is found in Acyrthosiphon pisum gut and honeydew and is known to attract aphid enemies like the hoverfly Episyrphus balteatus (De Geer)(Leroy et al. 2011). Both these bacteria were found only in one of the two studied aphid species. Sample and control consisted in 60mL of 868 culture medium (20 g of glucose and 10 g of both yeast extract and casein peptone per liter of distilled water) inoculated with S. xylosus and S. sciuri respectively and incubated for 2 days at 20°C.